1. Introduction
Travel reservation systems have been one of the most important tourism marketing channels [
1,
2]. From the perspective of tourists, and they can complete tour booking (a process that includes information search, booking, and paying) in the most convenient way, which helps realize personalized alternative selections and greater efficacy and efficiency [
3]. Furthermore, reservation behaviours can also ensure tourists’ satisfaction with their travel experiences [
4]. From the perspective of tourism and hospitality industry stakeholders, tourists’ reservation preferences can be understood, thus laying a foundation for the construction of marketing channels [
5]. In addition, an understanding of reservation can help managers develop more reasonable pricing strategies [
6]. Finally, online reviews have become an important source of information reflecting the tourists’ reservation experiences, which provides a reference for managers in improving the quality of travel services [
7].
Previous research on online travel reservation mainly focused on the hotel industry [
8,
9], lacking a focus on tourist attractions [
10,
11]. Tourists are the main users of reservation services for tourist attractions, so it is necessary to study their reservation intentions. On the one hand, it provides a wider perspective for travel reservation theories; on the other hand, it can help improve the reservation system of tourist attractions, and thus, enhance tourist satisfaction and safety.
Previous research on online travel reservations have used theories such as trust theory [
12], maturity theory [
13], information search theory [
14], self-efficacy theory [
15], etc. Among them, the TAM theory is widely applied to investigate online reservation intentions [
16,
17]. A well-designed, user-friendly reservation website enhances the online travel booking experience [
18]. It is clear that information technology not only facilitates the booking process but is also an important factor influencing reservation behaviour [
19]. In the actual reservation process for tourist attractions, tourists need to face a complex and diverse network and technological environment. At the same time, the COVID-19 pandemic has also promoted the digital transformation of tourist attractions [
20]. Tourist attractions in different tourism destinations have gradually reopened during the COVID-19 pandemic. Tourists prefer destinations close to home, especially short distances, and local attractions appear to be dominant in the recovery phase [
21]. On the premise of ensuring tourist safety, the government and tourism industry sectors have implemented a travel reservation and booking policy (i.e., ticket reservation, time-segment tour reservation, and visitor interval entry) to promote domestic tourism markets in China. According to statistics from the Ministry of Culture and Tourism of the People’s Republic of China, by the end of 2021, more than 6000 A-level attractions in China offered online reservation services. “No reservations, no travel” has been integrated into the travel life of residents. A total of 58.7 percent of respondents expressed that they often use online travel reservation platforms based on a special survey on tourist behaviour by the China Tourism Academy in 2021. Normalized and high-frequency reservations for tourist attractions have become the mainstream mode.
There exists the need to consider whether and how tourists perceive risks and external variables of government policy, which affect tourist reservation intentions within the COVID-19 context. Overall, considering the importance of tourists’ perspectives and the role of technology in the tourist attraction reservation process, the study introduced the technology acceptance model (TAM), which studies people’s willingness to use new technologies and explores the influencing factors of tourists’ reservation intentions of tourist attractions.
This study contributes to the body of knowledge about tourists’ reservation intentions in the COVID-19 context in two ways. First, this study reveals the antecedents that affect tourists’ reservation intentions. Second, this study extends TAM based on comprehensive insight into understanding tourists’ reservation intentions. The findings shed light on the theoretical investigation and sustainable development of reservation services.
5. Results
5.1. Nonresponse Bias and Common Method Bias
According to Armstrong and Overton’s suggestion [
86], SPSS 26.0 was used in this paper to conduct a non-response bias test of the questionnaire. First, the questionnaire was divided into two parts according to the time sequence of return: early responders (the first 25% of the questionnaires) and late responders (the last 25% of the returned questionnaires). Second, the two groups were compared by the chi-square test. The results showed that there were no significant differences in the control variables of gender between the two groups at the 5% confidence interval. Therefore, this study excluded the possibility of nonresponse bias.
In addition, Harman’s single-factor test was used to evaluate potential common method bias. All items are loaded into an exploratory factor analysis, the results of the non-rotating factor analysis are checked, and the minimum number of factors required to explain the variance of the variables is determined. When only one factor is extracted or it has strong explanatory power, it must be considered that there is a serious common method bias. According to the results of this study, the contribution rate of the general factors is not more than 50%, the first factor accounted for 27.8%, and the total contribution rate of the six factors is 65.7%. It can be seen that there is no common method bias.
5.2. Measurement Model
Measurement model evaluation usually included testing for reliability, convergent, and discriminant validity. Reliability assessment depends on Cronbach’s α and the composite reliability (CR) (see
Table 1), values of 0.7 to 0.9 are considered as satisfactory [
87].
The data show that the Cronbach’s α values of the six variables are all above 0.74. After each item was deleted, there was no significant improvement in the reliability of each scale. At the same time, all CR values are higher than 0.7. Therefore, the reliability of the questionnaire is very good, and the internal stability and consistency are high. These results suggest that the measurement model is reliable and valid.
Convergent validity is assessed using the average variance extracted (AVE) for each construct (see
Table 1). The AVE values of all constructs are between 0.37 and 0.612 in this study. Although the AVE value of perceived risk is less than 0.5, the composite reliability is higher than 0.6 and in the acceptable range of 0.36 to 0.5 [
88].
This study presents the results of discriminant validity assessment using Fornell-Larcker criterion [
88] and the heterotrait-monotrait ratio of correlations (HTMT) [
89]. The square root of each construct’s AVE is higher than the correlations with other constructs, so the Fornell-Larcker criterion was fulfilled (see
Table 2). HTMT is the ratio of the mean of indicator correlation between different constructs to the mean of indicator correlation between same constructs. As shown in
Table 3, the values of HTMT do not exceed the required threshold value of 0.90 by Gold et al. [
90]. These results suggest that discriminant validity is achieved.
Validity is the basis for measuring whether the item design is reasonable (see
Table 4). The overall KMO value was 0.859, greater than 0.6, and the
χ2 statistic test value was 2631.589 (
p < 0.001), which met the conditions of exploratory factor analysis. After the maximum variance orthogonal rotation of principal component analysis, it was found that there were 6 common factors with the eigenvalue of the questionnaire greater than 1, and the cumulative variance contribution rate was 65.66%, which was greater than 60%, which met the research requirements. The factor loadings of the 24 measurement items were all greater than 0.5, and they belonged to different dimensions, which were in line with the expected assumptions, indicating that the questionnaire design was reasonable.
5.3. Structural Model
Before hypothesis testing, the model fit indices’ ability to meet the requirements needs to be examined. In this paper, 11 indices such as
χ2/
df, absolute fit indices (GFI, AGFI, RMSEA) and value added fit indices (CFI, IFI, NFI, TLI), and parsimony corrected fit indices (PCFI, PNFI, PGFI) were selected to test the model fit (see
Table 5). The data show that, except for the three indices of GFI, AGFI, and NFI, all other indices have reached the reference standard in the initial structural model. To enhance the degree of fit between the theoretical model and the actual model of the sample, the initial model needs further modification [
91].
The modified model I takes the method of adding and subtracting observed variables. The normalized factor loading value should be greater than 0.50 and not greater than 0.95. Referring to this criterion, the observed variable PR4 should be deleted. After deletion, the
χ2/
df value of the model was changed, and other fit indices improved. The modified model II involved the method of revising the covariance of the residuals of the variables. Since the correlation between the variable residuals was not considered when constructing the theoretical model, the model fitting effect will be affected by the strongly correlated variable residuals. Referring to this standard and combining the correction indices provided by AMOS, the study established a correlation between the variable residuals with correlation and, after many operations, until the variable residuals were uncorrelated. The modified models’ fit indices were as follows in
Table 5. The data show that the
χ2/
df value of the model has changed, and all fit indices meet the reference standard.
According to the results, each model modification can reduce the
χ2/
df value, and other fit indices can be significantly improved. Therefore, these modifications are feasible in theory, and the model modification results are accepted. When another attempt was made to establish the connection between observed variables of different dimensions, it was found that the new structural relationship was not as ideal as the modified model II, so the modified model II was finally selected in this study (see
Figure 2).
The results of hypothesis testing in this study are shown in
Table 6.
Among the cognitive variables, perceived usefulness has a significantly positive effect on tourists’ reservation intentions for tourist attractions (β = 0.16, p = 0.03), while tourists’ perceived ease of use has a positive effect on perceived usefulness (β = 0.46, p < 0.001). Tourists’ reservation intention was also significantly affected by perceived risk (β = −0.32, p < 0.001). Thus, H1~H3 were supported.
Among the external variables, subjective norms have no significant effect on tourists’ reservation intentions (β = 0.07, p = 0.355). Government policy has a significantly positive effect on both perceived usefulness (β = 0.24, p = 0.003) and tourists’ reservation intentions (β = 0.47, p < 0.001). Thus, H4 was not supported, and H5~H6 were supported.
6. Discussion and Conclusions
6.1. Conclusions
This study takes the TAM as the theoretical basis to investigate the online reservation intentions for tourist attractions and its influencing factors. Two variables (perceived risk and government policy) were introduced to expand the theoretical model in the COVID-19 context.
An online survey was conducted in China and derived from a sample of 255 through the Questionnaire Star platform, the data for this research were analysed using SPSS 26.0 and AMOS 28.0. Then, this study analysed the influence of subjective norms, government policy, perceived usefulness, perceived ease of use, and perceived risk on reservation intention for tourist attractions.
Based on the above research, this paper draws the following conclusions: (1) subjective norms have no significant impact on reservation behaviour under voluntary situations; (2) perceived usefulness positively affects tourists’ reservation intentions; and (3) perceived risk has a significant negative impact on reservation intentions, and government policy is the main factor affecting tourists’ reservation intentions. Compared with perceived risk, the external variable of government policy has a greater impact on tourists’ reservation intentions.
6.2. Theoretical Implications
The primary objectives of this study are to identify tourists’ reservation intentions for tourist attractions in the COVID-19 context and to measure the influencing factors via the extended TAM. Specifically, the findings advance reservation services research in the following three ways.
First, this study contributes to the understanding of tourists’ intentions to reserve tourist attractions on theoretical grounds. While the development of hotel online booking is gaining popularity [
8,
26], previous studies on tourist attraction reservation is insufficient. This research fills this gap in the literature regarding which important aspects tourists consider when booking tourist attractions. Based on the TPB, the TAM is introduced in this study to analyse tourists’ intentions to reserve tourist attractions. Obviously, the development of information technology has become the basic support for the implementation of reservation systems. Some prior studies have also analysed the effects of technological factors on tourists’ reservation preferences [
19,
29]. In the same way, these results show that tourists’ perceived usefulness of the reservation systems positively affects their reservation intention. Tourists’ perceived ease of use positively affects perceived usefulness. These findings revalidate the value of the TAM in the study of reservation intention and further support the research of Li and Zhang [
58].
Second, according to the rapid development of travel reservations in China and the change in tourists’ travel behaviour since COVID-19, the variables of perceived risk and government policy are integrated into the TAM. The extended model not only confirms the predictive role of risk perception on reservation intentions, but also effectively improves the explanatory ability of the model, and helps deepen the understanding of reservation intention. The significant impact of the risk variable on reservation intention has been verified. Perceived risk is commonly examined as one of the various determinants of travel reservation intentions that were affected by the pandemic [
42]. In the face of complex information, virtual networks, and the spread of the COVID-19 epidemic, tourists will inevitably feel the risks, such as personal information, time, money, public health, and other aspects. These situations lead to tourists’ concerns about the safety of tourist attractions and reservation services, which in turn, affects their intention to make reservations. Tourist attraction reservation is the management measure advocated by the Chinese government in the context of COVID-19. To a certain extent, tourists’ reservation has significantly promoted. Therefore, it is reasonable to introduce risk and policy variables in this study, and the results of the study also show that it is necessary to expand TAM.
Finally, this study found that the influence path of subjective norms is not supported, i.e., tourists’ subjective norms do not significantly affect their reservation intentions. Tourists’ reservation intentions are generated in a situation of voluntary use, so social pressure from surrounding people may not directly affect individual reservation intention. This result validates Venkatesh and Davis’s [
69] view that “subjective norms have no significant effect on intentions in voluntary situations”. Although in TPB [
92], subjective norms are factors that directly affect behavioural intentions, Davis [
33] did not use subjective norms in the original TAM. Mathieson’s [
93] study also showed that subjective norms do not have a significant impact on reservation intention.
6.3. Practical Implications
These results have important implications for tourist attraction managers. Through online reservation systems, the query and traceability of tourist information can be realized, which is an indispensable means for tourist attractions to ensure safe operation under the normalization of epidemic prevention and control. At the same time, the operators and managers of tourist attractions should continuously strengthen the functional construction of the reservation system to improve the perceived usefulness of tourists, so that tourists can reserve and purchase tickets reasonably according to the bookable volume of the destination before departure, accurately plan the route, and arrange the itinerary reasonably. In addition to providing online ticket reservation and time segment tour reservation services, tourist attractions also need to actively develop digital experience products and popularize intelligent services (such as electronic maps, route recommendations, voice guides, information inquiry, feedback, etc.). It is necessary to constantly optimize the interaction between tourist attractions and users, which improves tourists’ perceived ease of use. Measures that can be taken include improving the timeliness of information provision, attaching great importance to the personal experience of users, and reducing the cost of information search for tourists.
The findings suggest that perceived risk is a negative determinant of booking intentions. Tourist attraction managers should pay attention to the security of reservation systems to reduce the perceived risks for tourists. The personal privacy information of tourists should be guaranteed to eliminate the possible disclosure risk. On the tourism destination level, the government should continue to promote the convenience of the “reservation system” and formulate reservation regulations. Using big data, cloud computing, the Internet of Things, and other means to build a smart tourism system can promote reservation services ability for tourist attractions. At the same time, the government needs to expand channels for booking and cooperate with stakeholders, such as tourism enterprises, tourist attractions, tour leaders, and communities, to create a good reservation environment [
94,
95].
According to the findings, reservation systems have an impact on tourist decision making and behavioural intention which would aid in destination marketing. This study shows that destination marketing organisations (DMOs) and tourist attraction marketers should improve promotional materials and content of online reservation platforms to meet market expectations [
96]. Thus, during the time of the COVID-19 pandemic, tourists can use online reservation applications that allow them to easily and securely obtain destination information and compare products and prices, etc. [
97,
98].
6.4. Limitations and Future Research
Several limitations of this study should be acknowledged, which may provide guidance for future research. Firstly, due to the impact of COVID-19, this study used a convenience sample of tourists through the Questionnaire Star platform. It is necessary to conduct a face to face survey. Secondly, the respondents in this study were Chinese tourists only. However, the differences in perceived risk might be influenced by cultural context. Thus, future research could investigate tourists’ reservation intentions for tourist attractions in different cultural backgrounds.